fix: avoid suture.Supervisor being shutdown early

Previously the Supervisor was shutdonw as soon as the main context was
canceled. Which cause the managed services to fail during shutdown (in
"trapShutdownCtx()") as the Supervisor was gone already.

We now pass a separate Context to Supervisor.ServeBackground() to avoid
this.

Fixes: #2282

opencloud/pkg/runtime/service/service.go

@@ -433,7 +433,7 @@ func Start(ctx context.Context, o ...Option) error {
 	// go supervisor.Serve()
 	// because that will briefly create a race condition as it starts up, if you try to .Add() services immediately afterward.
 	// https://pkg.go.dev/github.com/thejerf/suture/v4@v4.0.0#Supervisor
-	go s.Supervisor.ServeBackground(ctx)
+	go s.Supervisor.ServeBackground(context.Background())
 
 	for i, service := range s.Services {
 		scheduleServiceTokens(s, service)

services/nats/pkg/command/server.go

@@ -91,7 +91,9 @@ func Server(cfg *config.Config) *cobra.Command {
 			gr.Add(runner.New(cfg.Service.Name+".svc", func() error {
 				return natsServer.ListenAndServe()
 			}, func() {
+				logger.Info().Msg("Gracefully shutting down the NATS server...")
 				natsServer.Shutdown()
+				logger.Info().Msg("NATS server shutdown")
 			}))
 
 			grResults := gr.Run(ctx)